I have a few questions about economics.
Is economics a science, and/or is there any real debate about whether it is?
Should it be considered a science?
Using a regression analysis, what would be a good r squared value to bolster an economic theory.
Thanks in advance
Economics is considered a social science.
WAG:
It can be a science, but whether there is any debate over this I have no idea. I don’t know if the terminology is correct, but one could say that (in my understanding) there are both applied and research economics; which would be parallel to say engineering and physics.
If one can develop a hypothesis for a mathematical model of economics and test it using the scientific process, and the subject is deep enough to allow for subtopics, specialties, and continued study for many years to come, then I see no reason why you wouldn’t call anything a science. I would probably consider someone like [this guy](Steven D. Levitt) as being a scientist, while as I wouldn’t necessarily consider this guy to be.
I was a literature major so…
No particular number. If I were investigating something really complex (say historical crime rates) with a simple model with one surprising variable (number of municipal bandstands), an R-squared of .35 might be pretty exciting if the t stats and coefficients looked good. In a simpler situation (substitution between different sorts of chicken feed), an R-squared of .7 might be pretty underwhelming.
Okay, this question came up because I believe my housemate, who has an economics degree, is full of shit. First, he says economics isn’t a science. Then he changed his stance, saying that there’s a lot of debate about whether it is a science. He justified his position be saying that a good economic study usually has a R-squared value of about .4, and that that correlation hardly passes the rigorous scientific standards imposed by other hard sciences. I called bullshit on his claim, but I wanted to make sure I wasn’t wrong in my stance. Could you elaborating on some of what you said.
Ha ha ha ha ha! Hoo hoo, hee hee hee. A hah. Ah. Heh.
Ha.
Ah, no. No it isn’t.
Ahhh. Whee!
There’s not much more to it than what was already said. Models of simple things tend to have high R[sup]2[/sup] values, whereas models of complex things don’t. The things studied in economics are more complex and harder to isolate than the things studied in hard sciences, so you’d expect the R[sup]2[/sup] values to be lower.
What is this supposed to mean?
Well, I don’t know if there is a debate about it but I was considering a economics degree and I noticed many school offer it as a BA or BS.
I don’t have an opinion either way. However, something that has been explained to me by a college professor is that you need to be able to perform the scientific method on a hypothesis or theory. A quick google search turns up others that agree with that opinion: http://www.freeinquiry.com/intro-to-sci.html
The thing about R^2 is that it is always increased by adding explanatory variables, even ones which are deemed nto statistically significant.
Social sciences can use numbers and statistics–it doesn’t mean they are sciences.
Does someone else what to trot out Kughn and take a paradigm-development-related stab at this? I’m too daft.
No?
All right, in a nutshell: some people believe that the hallmark of a true “science” is a relatively wide degree of agreement within the discipline about methods, about determining truth, about what are important questions for the field, and so on. While there are some exceptions (rancorous debates over things like string theory, maybe? I don’t know, I’m no physicist), generally in the sciences these things are better nailed down.
In the social sciences, there is more variation and more disagreement on such matters. Compared to the hard sciences and to life sciences, economics is a field with less consensus, so it looks more like other social sciences in this regard.
I apologize for grossly oversimplifying this. It’s been awhile.
So the responsible thing to do is report your adjusted R-squared, which corrects for the number of independent variables.
Partha DasGupta is a fellow of the Royal Society. I don’t think they elect non-scientists.
Holy moly! Ask him when an R[sup]2[/sup] of .4 became the cutoff for being a science. All that says is what proportion of the variance is due to the independent variables (IIRC…). In other words, the complaint is meaningless since that value doesn’t make any particular statement about how it’s interpreted. If an economist or a physicists were to say, “40% of the observed effect is due to the variables we’ve studied,” both are being equally scientific.
So, even if his claim is true, it is independent of whether economics is a science. Your roomate’s conclusion really says more about his understand of economics & science than it does about the discliplines themselves.
As for whether it’s a science, define science. I always thought science was trying to test theories empirically and extrapolate new hypotheses from what we’ve already tested. That’s what scientists do, ain’t it? (E.g., string theory.) Well, that’s what economists do. So my answer is, “yes.” I’ve come across Popper-heads who offer contrary opinions, but their arguments have never impressed me. I’ve read about how real people aren’t rational, but I’ve never been to verify this. Hell, there is even a field of classroom experiments used to demonstrate economic conclusions just like physicists and chemists use! You might find this little piece by Alvin Roth interesting in this context.
Neoclassical economics is built from simple axioms, with mathematical rigor, and has been tested quite a bit. It holds up pretty well. It’s a pretty good approximation and economists are working to make it better. That sounds like science to me.
What is the debate here? Is someone arguing that econcomics is a harder science than other social sciences (psychology, sociology …) and is more like chemistry or physics? Is someone arguing that economics doesn’t rise to the standard of other social sciences, and is more like the humanities (philosophy, literature, history …)?
Consensus is a better sign of disciplinary maturity than its status as a science. People have been doing physics for over 2000 years, and it has been 350 years since the Principia Mathematica alone. By these standards, economics is a spanking new science, so it is only expected that the same consensus has not yet been reached.
It does not help that performing economic “experiments” presents certain difficulties that do not typically trouble hard scientists.
No, it basically boils down to my roommate being full of shit. He first said economics wasn’t a science. Then he claimed it shouldn’t be, and that there’s a lot of serious debate in the academic community about whether it is or not. He evidence was the r-squared value things I was talking about before. I specifically said it was a social science in the minds of most people who matter, and that it fits the definition of a science. But he says it’s debatable, to which I replied that it was, in the same way evolution is. However, I figured, since he does have a degree in it, I could be wrong. So, here we are.
The whole argument was tangential a to discussion we were having about consumer goods. He’s a hippie, and thinks the world would be better if people didn’t buy consumer goods that they had no “need” for, or that had elevated prices due to them being seen as status symbols. By this, he means that people wouldn’t buy anything like designer clothes (or even extra clothes), luxury cars, or Starbucks coffee. I say that if everyone took this attitude, it would ruin our economy. He says that it won’t have an effect because people will donate their money to non-profit organizations instead of spending it on unnecessary things. I know this makes absolutely no sense, but he is convinced that the economy would be unaffected by people deciding not to buy things at all, and tries to pull rank by flaunting his economics degree.
Wait, can statistics find causality? I am ignorant, I know, but I thought stats was for collecting and analyzing data. You can’t, say, make a hypothesis that x causes y because of A, B, and C, and then use statistics to thoroughly claim one way or the other. Sure, you might have a really strong correlation, but that doesn’t precisely determine the underlying cause.
What am I missing?
Usually there is a theory behind the staitstical analysis, often it is a fairly complicated mathematical model in which if A then B kind of conclusions are reached. You can then see if the data supports the model, so to speak. One of the main prolems with economics is that the data is not generated (usually) in controlled experimental situations, which can be replicated, so the data is much messier and usually leads to less “strong” results.
Looks like we have a debate about this.
Moved from GQ to Great Debates.
samclem GQ moderator